LGG-16. Retrospective and integrative analyses of molecular characteristics and their specific imaging parameters in pediatric grade 1 gliomas.

Pediatric low-grade gliomas (PLGG) are the most common brain tumors diagnosed during childhood and represent a heterogeneous group associating variable molecular abnormalities. To go further and develop specific statistical patterns integrating altogether tumor molecular background, imaging features and patient outcome, a retrospective study was performed in a group of non-NF1 grade 1 PLGGs. So, 78 children, followed from 2004 to 2017, were retrospectively reported. Their therapeutic management comprised surgery or surgery plus chemotherapies (e.g., carboplatin and vincristine). In this population, radiological and molecular patterns were pooled statistically to be able to identify specific prognostic subgroups. Considering all 78 patients, 59 had only a surgical removal and 19 patients were treated with post-operative chemotherapy. Twelve progressions were reported in the partially resected and chemotherapeutic groups, whereas 4 deaths occurred only in the highly treated patients. As expected, PLGG with BRAF pV600E and/or CDKN2A loss exhibited poor outcomes. Nevertheless, we were able to evidence statistically significant associations between molecular characteristics and their specific imaging presentation on diagnostic and 24-week MRIs. In the chemo-treated patients, the molecularly good risk situations were significantly linked to the presence of a large tumor cyst at diagnosis (> 50% of the tumor volume (TV)) and the appearance during treatment of a higher cystic proportion that we called “cystic conversion”. So, additionally to the WHO prognostic molecular patterns, precise MRI characteristics (notably cystic part proportions onto TV) were associated to BRAF p.V600E PLGGs, as well as to the more frequent molecular KIAA1549-BRAF fused PLGGs. To summarize, absence on diagnostic MRI of cystic parts and/or cystic conversion at 6-month chemotherapy time point were significantly linked to a worst prognosis and response to treatment. In future studies, these imaging features might be considered as innovative prognostic markers to use during treatment.